Science Briefs

Cold Climates, Warm Climates: How Can We Tell Past Temperatures?

When we see pictures of dinosaurs basking in tropical heat or wooly mammoths shivering in an ice-covered tundra, how do we know that they lived in such climates? We can tell indirectly from sediments and deposits laid down during these periods. The presence of tropical plant and animal remains at the polar latitudes indicate that significantly warmer conditions must have existed as compared to today. Conversely, the absence of tree pollen in the tundra probably means that conditions were too cold for trees to grow.

For more quantitative information you have to look in the oceans, and in particular, at the deep sediments that lie on the bottom of the seas. There, a steady rain of shells from small, surface-dwelling animals falls continually, eventually building up hundreds of meters of sediment. These sediments preserve the shells of these small animals for millions of years, all the way back to the age of the dinosaurs, 65 million years ago.

Fig. 1: Photo of foraminifera

Figure 1: Magnified (400x) photograph of a left coiling N. pachyderma foraminifera. (Photo: Deep Sea Repository at Lamont-Doherty Earth Observatory, Columbia University.)

The most important of these animals, foraminifera (or forams for short), make their tiny shells from a form of calcium carbonate (CaCO3). This carbonate is found in many common geological features, such as the White Cliffs of Dover, which were once at the bottom of the sea.

What makes calcium carbonate important? The carbonate, originally dissolved in the oceans, contains oxygen, whose atoms exist in two naturally-occurring stable isotopes, 18O and 16O. The ratio of these two isotopes tells us about past temperatures. When the carbonate solidifies to form a shell, the isotopic ratio in the oxygen (written as δ18O) varies slightly depending on the temperature of the surrounding water. The change is only a tiny 0.2 parts per million decrease for each degree of temperature increase. Nevertheless, this is sufficient for us to be able to estimate the temperature of the water in which the forams lived millions of years ago. From this, we can see that temperatures in the Arctic Ocean were about 10-15°C warmer at the time of the dinosaurs than they are today!

There is a complication, however. The δ18O value in the shells depends critically on what the δ18O value was in the surrounding sea water (H2O), and that can be as variable as the temperature! This variability arises because when water evaporates, the lighter molecules of water (those with 16O atoms as compared to those with 18O) tend to evaporate first. Therefore, water vapor is more depleted (fewer H218O molecules) than the ocean from which it evaporates. Thus, the ocean has more 18O in places where lots of water evaporates (like the sub-tropics) and less where it rains a lot (like the mid-latitudes).

Similarly, when water vapor condenses (to make rain for instance), the heavier molecules (H218O) tend to condense and precipitate first. So, as water vapor makes its way poleward from the tropics, it gradually becomes more and more depleted in the heavier isotope. Consequently snow falling in Canada has much less H218O than rain falling in Florida. Changes in climate that alter the global patterns of evaporation or precipitation can therefore cause changes to the background δ18O ratio.

In addition, the great ice-sheets that once covered North America, consisting of snow falling in what is now Canada, were very depleted in 18O. Now, enough water was held in these ice sheets to reduce the global average sea level by about 120m. Furthermore, there was also enough depleted water trapped in the ice to increase the average isotopic content of the oceans. And so the first thing we see when we analyze the shells from the bottom of the ocean, is the waxing and waning of the great ice sheets over the last 3 million years (figure 2). The same pattern over the last 400,000 years can also be seen in the isotopes measured in ice cores drilled from the remaining ice sheets in Greenland and Antarctica.

Fig. 2: Istotopic ratio plot

Figure 2: The variations of 18O in carbonate averaged over a large number of different cores (in order to isolate a global signal) over the last 600,000 years. Most clear are the regular oscillations of the glacial ice volume which follows small changes to Earth's orbit around the Sun (Milankovitch forcing).

In consequence, the many records of δ18O in ocean sediments and in ice cores, contain information about the temperature, evaporation, rainfall, and indeed the amount of glacial ice — all of which are important to know if we are to understand the changes of climate in the Earth's history. Unfortunately, trying to disentangle these multiple effects is complicated since we have one measurement with many unknowns.

The paleoclimate group at GISS is working to try to decode these records using the latest generation of numerical models of the atmosphere and ocean circulation. In those models, we have included most of the physics necessary to simulate the distribution of δ18O in the oceans and the atmosphere. In addition, we have developed models of foram ecology that allow us to estimate at what depths in the ocean and at what season the carbonate forms on average.

This sequence of models allows us, for the first time, to map simulated climate changes directly from the model to the carbonate in the sediments — the actual data that paleoceanographers have measured. Initial experiments have focused upon the large climate changes that occurred during the melting of the ice sheets between 20,000 and 10,000 years ago. The closer the modeled changes match those seen in the sediments, the greater the confidence we have in the realism of our models.

While this new approach is unlikely to show that mammoths spent their time on the beach enjoying the sun, it may provide better understanding of the complicated sequence of events that marked the end of the ice age. It should shed light on the very rapid climate changes that have occurred in the North Atlantic and Europe at the end of the last ice age. Those particular changes have been associated with changes in the amount of heat carried poleward by the Gulf Stream. If we can understand that process, we may be better able to estimate the probability of its recurrence as a possible consequence of continued global warming.


Schmidt, G.A. 1999. Forward modeling and interpretation of carbonate proxy data using oxygen isotope tracers in a global ocean model. Paleoceanography 14, 482-497.


Please address all inquiries about this research to Dr. Gavin Schmidt.